Abstract

AI voice technologies transform the human voice, contorting embodied identity markers into commodified interactive interfaces. As users continuously engage with AI voices, iterative training cycles reinforce voiced simulations, enabling cloned voices to supersede the embodied voices from whom they originate. Building on commentary made by Leuenberger that links social development and voice, we examine how voice cloning objectifies the voice and produces a simulacra, blurring distinctions between the real and simulation. Through remediation, developers cultivate the illusion of a conversational presence while masking media and social biases.